复杂图象轮廓提取与识别技术研究

Study on Complex Image Contor Extraction and Recognition Techniques

本文以X线指关节骨图象轮廓提取和识别为典型,研究了具有乘性和加性噪声干扰的复杂图象轮廓提取与识别技术。文中系统地研究了典型的微机图象识别系统设计问题.这类问题通常首先作图象特征与识别参数描述;如果识别特征信息在轮廓线上,则要研究图象轮廓跟踪器设计与实现,然后,研究识别参数提取与识别器设计与实现.文中还深入地研究了轮廓线的视觉模型,根据视觉模型,提出了抗干扰的轮廓跟踪技术:先作梯度变换,作粗轮廓跟踪,最后作精细轮廓跟踪.在精细轮廓跟踪时,使用多个初始点解决了因轮廓多处断裂防碍轮廓线提取的困难.最后对所研究的技术进行了上机实测。测试结果十分好,并已成功地用于实际系统中.

Taking the contor extraction and recognition of X-ray phalanx joint image as an example, thecontor extraction and recognition techniques of the image with multiplicative and additive noise are studied in thispaper. The typical microcomputer image recognition system design is developed systematically and usually it is inthe steps: desciibe the image feature and defime the recognition paras; make design of the image Contor extractorand recognition paras; if the related information is on the contors, make design of the image contor extractor andits realization, then estimation of the recognition paras, The design of the recognition unit and its realizationThe visual model of the image contor is also studied in the paper and based on the model. the contor ex-traction techniques against the complex noise is developed. The techniques are basically as follows: firstly, thegradient transform, the rough contor tracking , and the fine contor tracking. In the fine contor tracking, severalinitial points of the tracking are set to solve the difficulty of possible breaks of the contor tracking. All thetechniques designed are tested on microcomputer system and used in the practical system, and the experimental re-sults are excellent.